1 Introduction

The core message of this manuscripts are as follows:

  • Resource use differs between H. curtus and H. schistosus
  • Variation in fishing intensity of trawlers and gillnets affects patterns of resource use

This file was last update on 2020-05-23

2 Varition in fishing intensity along the sindhudurg coast

2.1 Calculating effort with depth

Fishing intensity: Vessel effort = Number of days fishing x No. of hauls x Average time per haul is calculated for each boat on each day. This is then totaled for the sampling duration at each site.

  • Should fishing intensity be analysed seperately by fishing gear.
  • Should calculation of fishing intensity factor in Gears.
dep_int = effort%>% 
  filter(Fishing.Location != "")%>% #removing missing location data
  dplyr::select(Date, Gear.Type, Fishing.Location, Boat.name,
                No..of.Hauls, Average.Haul.Time..Hours., No..of.Days.fishing, Depth..wav.)%>%
  group_by(Gear.Type)%>%
  mutate(n.days = ifelse(is.na(No..of.Days.fishing), #filling in missing data with means
                         median(No..of.Days.fishing, na.rm = T), No..of.Days.fishing),
         n.hauls = ifelse(is.na(No..of.Hauls), 
                          mean(No..of.Hauls, na.rm = T), No..of.Hauls),
         haul.time = ifelse(is.na(Average.Haul.Time..Hours.), 
                            mean(Average.Haul.Time..Hours., na.rm = T), Average.Haul.Time..Hours.))%>%
  ungroup()%>%
  mutate(effort = 
           (n.days)*(n.hauls)*(haul.time),
         Depth.m = Depth..wav.*1.87)%>%
  dplyr::select(Date, Boat.name, Gear.Type, Fishing.Location, Depth.m, effort)%>%
  drop_na()
Var1 Freq
GillNet 163
Trawler 107

Variable type: numeric

skim_variable Gear.Type n_missing complete_rate mean sd p0 p25 p50 p75 p100 hist
effort GillNet 0 1 1.12 0.67 0.17 0.75 1.00 1 4 ▅▇▂▁▁
effort Trawler 0 1 17.33 7.92 4.00 12.00 17.38 18 48 ▅▇▂▁▁
estimate estimate1 estimate2 statistic p.value
-16.20868 1.120948 17.32963 -21.11231 0 
## [1] 3.235314

2.2 Mapping fishing intensity

How do we use location and depth data gathered from fisher surveys to map fishing intensity?

  • Nearest landmarks are geocoded from google maps API
  • Latitude is extracted from landmark geocode
  • Match landmark latitude and depth from survey to GEBCO data
  • Extract final longitude from GEBCO data

Are points accurately depicted?

Survey based map of fishing intensity needs to be verified with onboard observations and/or remote GPS data.

Gear.Type ext
GillNet 47
Trawler 94

3 DIfference in distribution and habitat use between H. curtus and H. shcistosus

3.1 Mapping sea snake distribution with fisheries survey data

#Calculating sea snake abundance

snakes_den = snakes%>%
  filter(Species == "Hydrophis schistosus" | Species == "Hydrophis curtus",
         Fishing.Location != "", 
         !is.na(Depth.Caught..m.))%>%
  dplyr::select(Species, Fishing.Location, Depth.Caught..m., Date, Boat.Name.Owner, 
         No..of.Hauls, Average.Haul.Duration..Hours., Gear.Type)%>%
  group_by(Gear.Type)%>%
  mutate(n.hauls = ifelse(is.na(No..of.Hauls), 
                         median(No..of.Hauls, na.rm = T), No..of.Hauls),
         haul.time = ifelse(is.na(Average.Haul.Duration..Hours.), 
                         median(Average.Haul.Duration..Hours., na.rm = T),
                         Average.Haul.Duration..Hours.))%>%
  group_by(Date, Boat.Name.Owner)%>%
  summarise(HC = sum(Species == "Hydrophis curtus"),
         HS = sum(Species == "Hydrophis schistosus"),
         effort = last(n.hauls*haul.time),
         Fishing.Location = last(Fishing.Location),
         Depth.Caught..m. = last(Depth.Caught..m.))%>%
  gather(c("HC", "HS"), value = n, key = "Species")%>%
  mutate(CPUE = n/effort)%>%
  drop_na()
Gear.Type n effort
GillNet 396 541.12
Rampan 60 214.90
Trawler 40 185.50
Data summary
Name Piped data
Number of rows 228
Number of columns 4
_______________________
Column type frequency:
numeric 2
________________________
Group variables None

Variable type: numeric

skim_variable n_missing complete_rate mean sd p0 p25 p50 p75 p100 hist
HC 0 1 0.38 1.90 0 0 0 0 14.84 ▇▁▁▁▁
HS 0 1 1.32 7.32 0 0 0 0 94.98 ▇▁▁▁▁
Species ext
HC 49
HS 51

  • Why does the join reduce from 101 -> 92
  • Why is H. curtus only found around and off Malvan?

Do H. schistosus and H.curtus partition the coastal habitat along the depth axis?

3.2 Summarising depth use by species

Variable type: numeric

skim_variable sp n_missing complete_rate mean sd p0 p25 p50 p75 p100 hist
mean.depth HC 0 1 -29.59 10.9 -47.48 -38.93 -28.36 -21.05 -8.26 ▇▆▇▇▃
mean.depth HS 0 1 -21.56 9.8 -39.54 -27.49 -21.12 -13.95 0.02 ▅▅▇▆▂

3.3 Spatial segregation by species

ttl.ext ovrlp rel.ovlp
73 27 0.369863

3.4 Modelling depth use by species

estimate estimate1 estimate2 statistic p.value d
-8.037095 -29.59214 -21.55504 -3.873995 0.0001957 0.7766168
sp term estimate std.error statistic p.value adj.r.squared
HS (Intercept) 1.2516543 0.1038657 12.050698 0.0000000 0.4047722
HS mean.depth -0.0253018 0.0035796 -7.068387 0.0000000 0.4047722
HC (Intercept) -0.2516543 0.1038657 -2.422881 0.0179507 0.4047722
HC mean.depth 0.0253018 0.0035796 7.068387 0.0000000 0.4047722

After controling for effect of gear type with a mixed effect model. Depth caught still varies significantly with species.

Hence, prefered habitat in terms of depth may vary bewtween H. curtus and H. schistosus

4 DIfference in diet between H. curtus and H. schsitosus

4.1 No. of samples collected by sea snake species

Species N n n_male n_female prop
Hydrophis curtus 179 36 18 18 0.2257251
Hydrophis schistosus 605 93 55 36 0.7629256

A very low proportion of inviduals had gut content present at the time of sampling. Sampling accross species and sexes may be adequate for comparison.

4.2 Proportion of unidentified specimens

N_sp N_fam unid_sp unid_fam
26 17 0.4031008 0.2945736

A large portion of gut content specimens were unidentifiable. This is an unavoidable consquence of VGCA.

4.3 Diverisity of prey families found in gut content

Snake.Species Prey.Species.Richness
Hydrophis curtus 11
Hydrophis schistosus 20
hs.D.Value hs.StdErr hc.D.Value hc.StdErr
6.889195 0 9.029676 0
Overlap
6
D
0.2063492 
## $Simlarity
## [1] 0.2550017
## 
## $StdErr
## [1] 0.14

4.4 List of sea snake prey species

Snake.Species Prey.Family Prey.Species Percentage
Hydrophis curtus 2
Hydrophis curtus Unidentified 1
Hydrophis curtus Carangidae Alepes sp. 2
Hydrophis curtus Carangidae Caranx heberi 2
Hydrophis curtus Clupeidae Sardinella longiceps 1
Hydrophis curtus Clupeidae Sardinella sp. 2
Hydrophis curtus Cynoglossidae 1
Hydrophis curtus Cynoglossidae Unidentified 1
Hydrophis curtus Engraulidae Thryssa dussumieri 1
Hydrophis curtus Engraulidae Unidentified 2
Hydrophis curtus Leiognathidae Leiognathus sp. 2
Hydrophis curtus Nemipteridae Nemepteris sp. 1
Hydrophis curtus Scombridae Rastrelliger kanagurta 2
Hydrophis curtus Serranidae Epinephelus diacanthus 1
Hydrophis curtus Tetraodontidae Lagocephalus inermis 1
Hydrophis curtus Unidentified Unidentified 13
Hydrophis schistosus 1
Hydrophis schistosus Unidentified 1
Hydrophis schistosus Ariidae Arius caelatus 1
Hydrophis schistosus Ariidae Arius maculatus 2
Hydrophis schistosus Ariidae Arius sp. 3
Hydrophis schistosus Ariidae Unidentified 2
Hydrophis schistosus Carangidae Megalapsis cordyla 1
Hydrophis schistosus Clupeidae Sardinella longiceps 1
Hydrophis schistosus Clupeidae Sardinella sp. 3
Hydrophis schistosus Clupeidae Unidentified 2
Hydrophis schistosus Leiognathidae Leiognathus brevirostris 1
Hydrophis schistosus Leiognathidae Unidentified 2
Hydrophis schistosus Pleuronectidae Unidentified 1
Hydrophis schistosus Plotosidae Plotosus lineatus 4
Hydrophis schistosus Plotosidae Plotosus sp. 2
Hydrophis schistosus Scombridae Rastrelliger kanagurta 1
Hydrophis schistosus Serranidae Epinephelus diacanthus 4
Hydrophis schistosus Serranidae Epinephelus sp. 1
Hydrophis schistosus Sillaginidae Sillago sihama 1
Hydrophis schistosus Synodontidae? Unidentified 2
Hydrophis schistosus Teraponidae Terapon puta 2
Hydrophis schistosus Teraponidae Terapon sp. 1
Hydrophis schistosus Teraponidae Terapon theraps 2
Hydrophis schistosus Tetraodontidae Arothron sp. 2
Hydrophis schistosus Tetraodontidae Chelonodon laticeps 4
Hydrophis schistosus Tetraodontidae Lagocephalus inermis 26
Hydrophis schistosus Tetraodontidae Unidentified 1
Hydrophis schistosus Unidentified Unidentified 20

4.5 Relative abundance of prey families in sea snake gut content

4.6 Diverisity of prey families found in gut content

Snake.Species Prey.Family.Richness
Hydrophis curtus 10
Hydrophis schistosus 13
hs.D.Value hs.StdErr hc.D.Value hc.StdErr
6.889195 0.7849444 9.029676 0.9356905
0 1 2
Hydrophis curtus 10 9.029676 8.344828
Hydrophis schistosus 13 6.889195 4.225309

4.7 Overlap in prey of H. curtus and H. schsitosus

Overlap
7 
## $Simlarity
## [1] 0.2550017
## 
## $StdErr
## [1] 0.1118219
term df SumOfSqs R2 statistic p.value
fam_simboo$Snake.Species 1 1.556349 0.0511228 3.771403 0.002
Residual 70 28.886984 0.9488772 NA NA
Total 71 30.443333 1.0000000 NA NA

4.8 Prey preference

Snake.Species Prey.Family per.F per.W per.N IRI rank
Hydrophis schistosus Tetraodontidae 34.246575 38.7307267 35.869565 2554.804519 1
Hydrophis curtus Unidentified 44.827586 15.2823920 40.625000 2506.193436 1
Hydrophis schistosus Unidentified 23.287671 12.9453244 21.739130 807.720182 2
Hydrophis curtus Clupeidae 6.896552 29.9003322 9.375000 270.864360 2
Hydrophis curtus Engraulidae 10.344828 11.9601329 9.375000 220.708271 3
Hydrophis curtus Carangidae 13.793103 3.3222591 12.500000 218.238057 4
Hydrophis schistosus Ariidae 9.589041 8.8407094 8.695652 168.156891 3
Hydrophis schistosus Clupeidae 6.849315 10.9456402 6.521739 119.639584 4
Hydrophis schistosus Plotosidae 8.219178 7.7882440 6.521739 117.616299 5
Hydrophis curtus Scombridae 6.896552 6.6445183 6.250000 88.927712 5
Hydrophis schistosus Teraponidae 6.849315 6.5673841 5.434783 82.206621 6
Hydrophis schistosus Serranidae 5.479452 7.6303742 5.434783 71.589900 7
Hydrophis curtus Cynoglossidae 6.896552 3.9867110 6.250000 70.598007 6
Hydrophis curtus Serranidae 3.448276 16.6112957 3.125000 68.056192 7
Hydrophis curtus Leiognathidae 6.896552 2.9900332 6.250000 63.724367 8
Hydrophis curtus Tetraodontidae 3.448276 7.6411960 3.125000 37.124814 9
Hydrophis schistosus Leiognathidae 4.109589 1.0524654 3.260870 17.726034 8
Hydrophis curtus Nemipteridae 3.448276 1.6611296 3.125000 16.503895 10
Hydrophis schistosus Synodontidae? 2.739726 1.1840236 2.173913 9.199826 9
Hydrophis schistosus Sillaginidae 1.369863 2.4206704 1.086957 4.804968 10
Hydrophis schistosus Scombridae 1.369863 1.2103352 1.086957 3.146975 11
Hydrophis schistosus Pleuronectidae 1.369863 0.3683629 1.086957 1.993588 12
Hydrophis schistosus Carangidae 1.369863 0.3157396 1.086957 1.921502 13

4.9 Selectivity of prey sizes

Variable type: numeric

Snake.Species n_missing complete_rate mean sd p0 p25 p50 p75 p100
Hydrophis curtus 1 0.86 2.59 0.57 2.00 2.18 2.45 2.88 3.5
Hydrophis schistosus 0 1.00 3.59 1.32 1.06 3.35 3.80 4.30 5.5
estimate estimate1 estimate2 statistic p.value d
-0.6961959 2.76775 3.463946 -2.386272 0.0237889 0.6420566

5 Difference in resouce use betweek H. curtus and H. schistosus?

Carbon and Nitrogen isotope ratios were compared accorss species. Plasma and scale samples were used to compare short term and long term resource use respectively. Multiples metrics including niche width (SEA), variance (range), overlap (%) were used.

5.1 Number of samples analysed

Species Plasma Scales
Hydrophis curtus 6 12
Hydrophis schistosus 27 25

5.2 Summary statistics on Carbon and Nitrogen stable isotopes

Species Tissue.type Delta.Carbon Delta.Nitrogen
Hydrophis curtus Plasma -17.6±0.6 13.23±0.56
Hydrophis curtus Scales -15.44±0.59 14.57±1.26
Hydrophis schistosus Plasma -16.72±1.14 14.45±1.12
Hydrophis schistosus Scales -14.95±1.25 15.13±1.14

5.3 Difference in niche width between sea snakes

5.3.1 Maximum likelihood estimate of SEA

TA SEA SEAc
Hydrophis curtus.Plasma 0.74470 0.6368364 0.8491152
Hydrophis curtus.Scales 4.58370 2.4210563 2.7236883
Hydrophis schistosus.Plasma 10.19485 3.1208774 3.2851342
Hydrophis schistosus.Scales 20.43275 4.4823982 4.7065181

As maximum likelihood can only porivde point estimates of SEA, a bayesian model was used to provide more robust comparison of niche width.

5.3.2 Bayesian estimate of SEA

Species Tissue SEA.mean SEA.sd SEA.se
Hydrophis curtus Plasma 1.107423 0.6453675 0.0102042
Hydrophis curtus Scales 4.698527 1.0578248 0.0167257
Hydrophis schistosus Plasma 2.720573 0.9673557 0.0152952
Hydrophis schistosus Scales 3.304453 0.7527456 0.0119020

5.3.3 Testing difference in species niche area by tissue type

Tissue p
Plasma 0.94925
Scales 0.13025

Hyp: H. schistosus SEA is larger than H. curtus

While niche width is slightly larger in H. schistosus, it is not significantly different. The posterior distribution of SEA^b has right skewed long tail (masked by the limits of the graph for visual clarity) possibly due to the low sample size. Niche width doesn’t seem to change accross different tissue types either, indicating stability over variying periods of assimilation.

5.4 Visualising posterior ellipses to compare species isotopic niche

As seen in the SEA estimation, low sample size for H. curtus causes greater uncertainty in the estimation of standar ellipses even with bayesian inference. Both carbon and nitrogen isotpes seem to be enriched in scales when compared to plasma. Degree of overlap appears to be low for both tissues.

  • Samples sizes for H. curtus in both tissues need to be increased despite using bayesian methods
  • Unusual outlier in H. schistosus scales data needs to be checked

5.5 Relative overlap in bayesian standard ellipses

Turnover Avg.overlap sd
Long-term 0.4244165 0.0827186
Short-term 0.0846205 0.0734810

Overlap between H. curtus and H. schistosus seems to be high in when comparing long term resource use, i.e., scales.

However, overlap between H. curtus and H. schistosus seems to be very low when comparing short term resource use, i.e., plasma.

This difference could be caused by:

  • low sample size for H. curtus
  • differences in processing of scales and plasma, i.e., lipid extraction

6 Does fishing intensity affect the distribution of H. curtus and H. schistosus?

6.1 Spatial overlap between fisheries and sea snakes

gear sp overlap extent.sp rel.ovlp
GillNet HC 30 49 0.6122449
GillNet HS 39 51 0.7647059
Trawler HC 47 49 0.9591837
Trawler HS 46 51 0.9019608 
## Error in (function (classes, fdef, mtable) : unable to find an inherited method for function 'select' for signature '"grouped_df"'

6.2 Variation in relative proportion of HC with fishing intensity

term estimate std.error statistic p.value
(Intercept) -0.2644395 0.1108086 -2.3864521 0.0197591
GillNet 0.0030194 0.0062982 0.4794069 0.6331652
Trawler -0.0004375 0.0012801 -0.3417601 0.7335702
mean.depth -0.0258173 0.0037580 -6.8699998 0.0000000
adj.r.squared AIC
0.3899049 65.93959
Sum Sq Df F value Pr(>F)
GillNet 0.0306342 1 0.2298310 0.6331652
Trawler 0.0155683 1 0.1167999 0.7335702
mean.depth 6.2908730 1 47.1968975 0.0000000
Residuals 9.1970078 69 NA NA

7 Overlap between fisheries catch and sea snake diet

7.1 Sampling Adequacy

Gear.Type N Haul.Hours
Gill Net 38 35.41667
Trawler 140 434.65000

7.2 Catch weight

## 
##  Shapiro-Wilk normality test
## 
## data:  log(tonnage$Total.Catch..kg.)
## W = 0.96656, p-value = 0.02034
estimate estimate1 estimate2 statistic p.value parameter conf.low conf.high method alternative
-0.8347031 4.585903 5.420606 -3.896286 0.0002104 75.08903 -1.261464 -0.4079425 Welch Two Sample t-test two.sided 
## [1] 0.8447173

7.3 Diverisity of prey families found in fisheries catch

Gear.Type Family.Richness
Gill Net 15
Trawler 48

7.4 No. of Sea snake prey families found in fish catch and overlap

Gear.Type snake Nfam
Gill Net HC 7
Gill Net HS 6
Trawler HC 9
Trawler HS 9
Gear.Type snake Mean.prop
Gill Net HC 0.8204191
Gill Net HS 0.7958067
Trawler HC 0.7356892
Trawler HS 0.4831901
Gear.Type estimate HC HS statistic p.value d
Gill Net 0.0246124 0.8204191 0.7958067 0.6314888 0.5296802 0.1448735
Trawler 0.2524990 0.7356892 0.4831901 7.0551602 0.0000000 0.8924545
snake estimate GN TR statistic p.value d
HC 0.0847299 0.8204191 0.7356892 2.469163 0.0153629 0.3588295
HS 0.3126166 0.7958067 0.4831901 7.761277 0.0000000 1.0913295

7.5 Sea snake prey species in fisheries catch

Species constituting >10% of the catch on average are represented

8 Does isotopic niche of sea snakes change with local fishing intensity?

As it is infeasible to to find sites with little to no fishing pressure on mainland India. The ultimate objective of this project remains unfulfilled it is nigh impossible to conduct a study with controls and test sites with varying levels of fishing pressure.

  • How do you test the niche shift hypotheses with out spatial replicates?

    While the control - test study design failed, fishing effort data was collected from multiple vessels (Gillnet and Trawlers) landing at the Malvan harbour over the period of 1.5 years along side diet, abundance (snakes and prey), and isotopic data. So we ask the following questions:

    • Is there a corellation between fishing intensity and isotopic ratios?
    • Do niche metrics (width, overlap) vary with fishing pressure (high, medium and low preiods during the year)?

    What temporal resolution should be used for the analysis?

    • Day, week, month or season.

  • Which is appropriate and why?

    Depending on the tissue being analysed. Scales - month, Plasma - week

  • What are the existing data and sampling structures and what can be improved?

#Stable isotope data

sia = read.csv("../Data/Stable Isotope Data_CEAS_241119.csv")

#joining sia data to snake data

sia_snakes = sia%>%
  filter(Tissue.type != "Gut Content")%>%
  left_join(snakes, 'Field.Code')%>%
  mutate(Lab = "CEAS")%>%
  rename(Delta.Carbon = d13C..vpdb. , Delta.Nitrogen = d15N..N2.air.)%>%
  dplyr::select(Date, Field.Code, Species, Snout.to.Vent..cm., Sex, Gravid, Class,
                Gear.Type, Fishing.Location, Depth.Caught..m.,
                Plasma.Color, Delta.Carbon, Delta.Nitrogen, Lab, 
                Tissue.type, Month, Year)

sia_li <- sia_fi%>%
  group_by(Field.Code, Species, Tissue.type, Delta.Carbon, Delta.Nitrogen)%>%
  nest()%>%
  mutate(cell = map(data, cell.ext),
         celldf = map(cell, as.data.frame))%>%
  dplyr::select(celldf)%>%
  unnest()%>%
  inner_join(fi, by = c("x", "y"))

write.csv(sia_li, "./Data/SIA_fishing intensity.csv")

8.1 Sample size

##                       
##                        Plasma Scales
##   Hydrophis schistosus     11     12

8.2 Variation Carbon isotope ratio with fishing intensity

8.3 Modeling isotope ratios with fishing intensity

8.3.1 Continous model

Species Isotope term estimate std.error statistic p.value adj.r.squared
Hydrophis schistosus Delta.Carbon (Intercept) -15.4011808 0.6232436 -24.711336 0.0000000 0.1925198
Hydrophis schistosus Delta.Carbon intensity -0.0119892 0.0065172 -1.839621 0.0989712 0.1925198
Hydrophis schistosus Delta.Nitrogen (Intercept) 15.7569605 0.3712895 42.438473 0.0000000 0.2023267
Hydrophis schistosus Delta.Nitrogen intensity -0.0073013 0.0038825 -1.880548 0.0927251 0.2023267

Samples sizes for H. curtus are low, however, fishing intensity seems to have a significant depletion effect on plasma isotope ratios in H. schistosus.

8.4 Difference in niche width between sea snakes

8.4.1 Maximum likelihood estimate of SEA

TA SEA SEAc
Hydrophis schistosus.Low 2.1424 1.234995 1.440828
Hydrophis schistosus.High 1.2033 2.182545 4.365089

As maximum likelihood can only porivde point estimates of SEA, a bayesian model was used to provide more robust comparison of niche width.

8.4.2 Bayesian estimate of SEA

Species FI.Class SEA.mean SEA.sd SEA.se
Hydrophis schistosus High 3.468871 3.1500851 0.0498072
Hydrophis schistosus Low 1.529057 0.6105229 0.0096532

8.4.3 Testing difference in species niche area by tissue type

Species p
Hydrophis schistosus 0.82375